Mixed model

This model is trained with 50% null plots and 50% not null plots. In not null plots, there are three different violations namely non-linearity, heteroskedasticiy and non-normality.

The following is its performance on the test set.

Metric Estimator Estimate
accuracy binary 0.8362288
Not null Null
Not null 5469 708
Null 1611 6372
Violation Accuracy Correct Total
heter 0.7283333 1311 1800
non_normal 0.8172619 2746 3360
null 0.9000000 6372 7080
poly 0.7354167 1412 1920

The model predicts the following plot to be not null with \(\hat{p} = 55\%\).

The model predicts the following plots (\(b = 64\)) with \(\hat{p} < 50\%\).

## [[1]]
## Image. Width: 420 pix Height: 525 pix Depth: 1 Colour channels: 3 
## 
## [[2]]
## Image. Width: 420 pix Height: 525 pix Depth: 1 Colour channels: 3 
## 
## [[3]]
## Image. Width: 420 pix Height: 525 pix Depth: 1 Colour channels: 3 
## 
## [[4]]
## Image. Width: 420 pix Height: 525 pix Depth: 1 Colour channels: 3 
## 
## [[5]]
## Image. Width: 420 pix Height: 525 pix Depth: 1 Colour channels: 3 
## 
## [[6]]
## Image. Width: 420 pix Height: 525 pix Depth: 1 Colour channels: 3 
## 
## [[7]]
## Image. Width: 420 pix Height: 525 pix Depth: 1 Colour channels: 3 
## 
## [[8]]
## Image. Width: 420 pix Height: 525 pix Depth: 1 Colour channels: 3 
## 
## [[9]]
## Image. Width: 420 pix Height: 525 pix Depth: 1 Colour channels: 3

Visual experiments

Metric Estimator Estimate
accuracy binary 0.755102
Not null Null
Not null 444 0
Null 144 0
Metric Estimator Estimate
accuracy binary 0.7142857
Not null Null
Not null 420 0
Null 168 0

## [[1]]
## Image. Width: 420 pix Height: 525 pix Depth: 1 Colour channels: 3 
## 
## [[2]]
## Image. Width: 420 pix Height: 525 pix Depth: 1 Colour channels: 3 
## 
## [[3]]
## Image. Width: 420 pix Height: 525 pix Depth: 1 Colour channels: 3 
## 
## [[4]]
## Image. Width: 420 pix Height: 525 pix Depth: 1 Colour channels: 3 
## 
## [[5]]
## Image. Width: 420 pix Height: 525 pix Depth: 1 Colour channels: 3